Archives of Environmental & Occupational Health
ISSN: 1933-8244 (Print) 2154-4700 (Online) Journal homepage: http://www.tandfonline.com/loi/vaeh20
Biomarkers of Insecticide Exposure and Asthma in Children: A National Health and Nutrition Examination Survey (NHANES) 1999–2008 Analysis M. E. Perla, Tessa Rue, Allen Cheadle, James Krieger & Catherine J. Karr To cite this article: M. E. Perla, Tessa Rue, Allen Cheadle, James Krieger & Catherine J. Karr (2015) Biomarkers of Insecticide Exposure and Asthma in Children: A National Health and Nutrition Examination Survey (NHANES) 1999–2008 Analysis, Archives of Environmental & Occupational Health, 70:6, 309-322, DOI: 10.1080/19338244.2014.910490 To link to this article: http://dx.doi.org/10.1080/19338244.2014.910490
Accepted author version posted online: 22 Aug 2014.
Submit your article to this journal
Article views: 177
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=vaeh20 Download by: [Australian National University]
Date: 07 November 2015, At: 22:14
Archives of Environmental & Occupational Health (2015) 70, 309–322 C Taylor & Francis Group, LLC Copyright ISSN: 1933-8244 print / 2154-4700 online DOI: 10.1080/19338244.2014.910490
Biomarkers of Insecticide Exposure and Asthma in Children: A National Health and Nutrition Examination Survey (NHANES) 1999–2008 Analysis M. E. PERLA1, TESSA RUE2, ALLEN CHEADLE3, JAMES KRIEGER4, and CATHERINE J. KARR5
Archives of Environmental & Occupational Health 2015.70:309-322.
1
Northwest Pediatric Environmental Health Specialty Unit, School of Public Health, University of Washington, Seattle, Washington, USA 2 The Institute of Translational Health Sciences, University of Washington, Seattle, Washington, USA 3 Department of Biostatisticis, Center for Biomedical Statistics, School of Public Health, University of Washington, Seattle, Washington, USA 4 Public Health Department, Seattle and King County, Seattle, Washington, USA 5 Department of Pediatrics, University of Washington, Seattle, Washington, USA Received 29 October 2013, Accepted 21 March 2014
Pesticide exposure is a potential risk factor for increased asthma prevalence among children. The authors used National Health and Nutrition Examination Survey (1999–2008) biomarker data to evaluate dialkylphosphate (DAP) urinary concentrations, serum dichlorodiphenyldichloroethylene (DDE), and asthma among school-aged children (Mexican American, Non-Hispanic Black, NonHispanic White). Poisson logistic regression included age, sex, nativity, poverty index ratio, tobacco smoke exposure, and body mass index covariates. No association was found between DAP (N = 2,777) and asthma outcomes; adverse effect of DDE (N = 940) was suggested for Current Wheeze. Subgroup analyses identified positive associations with some asthma outcomes among Non-Hispanic Blacks, whereas inverse associations were identified among Mexican Americans. Results support previous associations observed among children’s DDE exposure and wheeze. Characterization of risk factors for pesticide exposure and disease recognition among Mexican Americans is needed. Keywords: asthma, children, DDE, epidemiology, pesticides
Introduction There has been an increase in the asthma prevalence rate among US children during the past 30 years.1–3 Children have a higher asthma prevalence rate than adults, and low-income racial and ethnic groups continue to be disproportionately affected.4–6 The causes of asthma and the reasons for its increased occurrence have remained little understood7; however, a number of well-established environmental risk factors have been identified for its exacerbation in individuals. The role of nonpersistent and bioaccumulative pesticide exposures in the development of asthma and the exacerbation of this disease has received attention due to the widespread and continued use of pesticides in both urban8–12 and rural13,14 environments and biological plausibility.15 Few epidemiological studies have explored the relationship between specific pesticide chemical types and asthma or asthma-related symptoms
in children. Prospective studies have found associations between prenatal and breast milk exposures to organochlorine pesticides and asthma, wheeze, and lower respiratory tract infections in newborn and young children in Spain.16–19 In the Ontario Farm Family Health Study, researchers did not observe clear associations between self-reported maternal pesticide exposures during pregnancy and asthma outcomes of their offspring.20 Studies of adults in agricultural settings, in contrast, have identified links between self-reported lifetime or current exposure to pesticides and respiratory health.21–23 We explored the hypothesis that pesticides are associated with asthma in school-aged children 7–16 years of age using population-based data of biomarker concentrations. Given the disparities in asthma outcomes and exposure to pesticides among subpopulations, effects by country of birth and race/ethnicity were also examined.
Methods Address correspondence to M. E. Perla, PhD, MS, Northwest Pediatric Environmental Health Specialty Unit, School of Public Health, University of Washington, Box 359739, 325 9th Avenue, Seattle, WA 98104, USA. E-mail: [email protected]
Data Source We obtained demographic, asthma, early childhood health, and chemical biomarker data collected by the National
310 Health and Nutritional Examination Surveys (NHANES). NHANES is conducted following a complex, multistage area sampling protocol to generate cross-sectional representative data of the civilian noninstitutionalized population living in the United States. To increase sample sizes, we merged several NHANES survey cycle data sets from 1999 to 2008. During the greater part of this time frame, Mexican Americans, non-Hispanic African Americans, low-income Non-Hispanic White Americans, and adolescents were oversampled for the purpose of generating sufficient sample sizes to derive reliable population-based estimates.24
Archives of Environmental & Occupational Health 2015.70:309-322.
Asthma and Asthma Symptoms Recognizing that the likelihood of a clinical diagnosis or selfrecognition of asthma may vary among subpopulations,25–27 we examined 3 dichotomous asthma outcome variables derived from NHANES medical examination and respiratory health survey questionnaires: (1) “Has a doctor or other health professional ever told you that you have asthma?” (Ever Asthma); (2) “Do you still have asthma?” (Current Asthma); and (3) “In the past 12 months have you had wheezing or whistling in your chest?” (Current Wheeze).
Biomarkers of Exposure The primary independent variables of interest were (1) common metabolites of organophosphate insecticides (dialkyl phosphates), a nonpersistent insecticide, and (2) dichlorodiphenyldichloroethylene (p,p’-DDE), the major metabolite of the organochlorine insecticide dichlorodiphenyltrichloroethane (DDT). NHANES participants 6 years of age and older submitted a whole-blood specimen and spot urine specimen. From the blood specimen, cotinine serum levels, a metabolite of tobacco smoke exposure and covariate of interest, were measured for all participants. A randomly selected subset of participants 6 years and older had urine specimens evaluated for concentrations of the 6 possible dialkylphosphate (DAP) metabolites of organophosphate pesticides: dimethylphosphate (DMP), dimethylthiophosphate (DMTP), dimethyldithiophosphate (DMDTP), diethylphosphate (DEP), diethylthiophosphate (DETP), and diethyldithiophosphate (DEDTP). For the analysis of DAP, we summed the 6 organophosphate metabolite concentrations after converting metabolites to nanomolar units by dividing the analyte concentrations by their molecular weight and multiplying by 1000. We adjusted the estimates of urinary biomarkers by correcting each organophosphate metabolite for urinary creatinine before conversion and DAP aggregation.28,29 Because specific organophosphate urinary metabolites had different percentages of observations above the limit of detection, ranging from 29% to 70%, we summed dimethyl alkylphosphate pesticides and diethyl alkylphosphate separately to yield DMAP and DEAP concentrations, for sensitivity analyses. For a separate, randomly selected subset of participants 12 years of age and older, DDE was measured in serum.
Perla et al. Nearly 100% of the DDE observations were above the limit of detection.
Covariates Our a priori selection of available NHANES covariates for descriptive statistics focused on risk factors associated with pesticide exposure and asthma or its diagnosis. These covariates included demographics and socioeconomic status (age, sex, race/ethnicity, country of birth, poverty index ratio [PIR], health insurance coverage), body mass index (BMI) by age, and exposure to tobacco smoke (prenatal smoking, cotinine). We examined data of participants 6–11 and 12–15 years of age available in the DAP data set, and data of adolescents 12–15 years of age available in the DDE data set. To evaluate disparities by race/ethnicity, we conducted stratified analyses of Non-Hispanic Whites, Mexican Americans, and NonHispanic Blacks. Although NHANES collects data from other racial/ethnic groups including Other Latinos (who do not have Mexican American heritage but who identify as Hispanic or Latino), these groups could not be considered for analysis due to sample size considerations. We dichotomized country of birth (US-born, foreign-born) and health insurance coverage (yes, no). The PIR was dichotomized to generally represent families meeting or not meeting poverty eligibility guidelines for federal programs: ie, PIR 10%) of this cross-sectional study prompted use of Poisson logistic regression with robust variance estimation.31 To select exposure percentile cutpoints reflective of changes in risk of asthma with increasing pesticide exposure levels, we used linear splines to model the shape of the dose-response trend.32 Ever Asthma had the highest prevalence rate and overlaps with Current Asthma and Current Wheeze; therefore, cutpoints identified for Ever Asthma were used to evaluate all health outcomes. This process identified 25th percentile as cutpoint for creatinine-corrected DAP and the 40th and 80th percentiles as biomarker cutpoints for DDE. We used these cutpoints to define levels of exposure for each pesticide.
Archives of Environmental & Occupational Health 2015.70:309-322.
Archives of Environmental & Occupational Health We estimated the unadjusted and adjusted asthma prevalence ratios for increasing exposure levels using the 25th percentile for DAP and the 40th percentile for DDE as the referent categories. Adjusted prevalence rate ratio (PRR) models for the DAP analysis included sex, race/ethnicity, BMI (continuous), PIR prenatal smoking, and log cotinine (continuous). As the DDE analysis had a smaller sample sizes and number of cases, for these models we reduced the number of covariates and adjusted for sex, BMI and the poverty index ratio. Due to the smaller race/ethnicity-stratified sample sizes, particularly for the DDE data set, we further reduced the number of covariates by using bivariate regression to select variables with the greatest influence in our sample. The stratified DAP models were adjusted for sex, cotinine, and BMI; the stratified analysis of DDE included BMI as a covariate. Statistical tests were 2-sided and our threshold for statistical significance was p < .025 and for marginal significance .025 ≤ p ≤ .10. The geometric mean of DAP in the 1999–2001 cohort of participants was significantly higher than in the larger analytical sample, so we evaluated the influence of results specific to survey cycle. We assessed the impact of tobacco smoke exposure on our results by removing participants with high cotinine exposure levels indicative of an active smoker.33 In addition, for DAP metabolites, we evaluated the potential impact of fasting time, which may influence the individual measurements of creatinine. Because we could not include country of birth in the stratified multivariate analysis, we conducted separate analyses restricted to US-born participants. This effort focused on Mexican Americans, since they had the largest percentage of foreign-born participants in the DAP (20%) and the DDE (25%) pesticide data sets, whereas Non-Hispanic Blacks and Non-Hispanic Whites had roughly 3% foreign-born in both pesticide data sets. For these analyses, only BMI was included as a covariate. We used Stata 1034 for our analysis. NHANES survey sample weights were used in estimation of unadjusted and adjusted geometric means, prevalence rate ratios, standard error estimates, and 95% confidence intervals, as well as exposure percentiles.
Results NHANES participants from 1999 to 2008 included 10,077 6–15-year-old Non-Hispanic Whites, Mexican Americans, and Non-Hispanic Blacks (Table 1). Thirty-six percent (3,622) 6–15-year-old participants submitted urine samples for measurement of DAP concentrations. With the exception of oversampling differences of adolescents, individuals who contributed urine samples for DAP analysis were similar to the larger NHANES participant group. During the 1999–2004 time frame, 962 participants aged 12–15 years were randomly selected to submit a blood sample for evaluation of DDE. For the analysis of the relationship between pesticides and asthma, the DAP data set was reduced by 845 (23%), primarily due to the incomplete analysis of all 6 dialkylphosphate metabolites needed to generate a DAP vari-
311 able. The DDE sample size was reduced by 22 (2%), primarily due to missing prenatal smoking and BMI observations. In both pesticide data sets, Non-Hispanic Blacks had the highest prevalence rates of Ever Asthma and Current Asthma compared with Non-Hispanic Whites and Mexican Americans (Table 2). With the exception of Current Wheeze in the DDE data set, Non-Hispanic Blanks also had the highest estimates of Current Wheeze. There were some statistical differences of estimated uncorrected geometric mean DAP pesticide concentrations by race ethnicity; however, there were no differences in the creatine-corrected estimates (Table 2). The DDE geometric mean among Mexican Americans was roughly 3 times higher than those of Non-Hispanic Whites and Non-Hispanic Blacks. Mexican Americans also had the lowest geometric mean levels of cotinine in both data sets: NonHispanic Whites had cotinine levels roughly 3 times higher than Mexican Americans, and Non-Hispanic Blacks had estimates 4 times higher. Unadjusted and adjusted asthma outcome prevalence rate ratios by increasing urinary DAP concentration percentiles were largely null, with point estimates generally greater than 1.0 for all participants (Tables 3a and 3b). Sensitivity analyses of DMAP and DEAP revealed similar null results (data in Appendices A and B). In stratified analysis, among NonHispanic Blacks, a positive trend was suggested for Current Wheeze among 6–11-year-old and Ever and Current Asthma among 12–15-year-olds. In contrast, an inverse trend was suggested for Ever and Current Asthma and Current Wheeze among Mexican American adolescents. Increasing percentiles of serum DDE were associated with increased Current Wheeze but not Ever or Current Asthma (Table 4). Subgroup analyses were limited by small sample size but supported this association among Non-Hispanic Whites and Non-Hispanic Blacks. In contrast, an inverse effect among Mexican Americans was shown for both Current Asthma and Current Wheeze. Our sensitivity analyses addressing survey cycle influence, number of hours fasted since specimen collection, and removing individuals with high body burden of cotinine did not alter the patterns reported in Tables 3a and 3b. Additional sensitivity analyses focused on US-born Mexican Americans. Increasing DAP and of DDE exposure levels did not differ from our core analyses (data available in Appendices A and B).
Comment Overall, this study revealed no clear associations between asthma and biomarkers of DAP or DDT insecticides among school-aged children in the United States. Although the analyses of asthma outcomes and urinary metabolites of DAP were largely null overall, subgroup analyses demonstrated marginally significant positive trends for Non-Hispanic Blacks for Current Wheeze among 6–11-yearolds, and Ever and Current Asthma among adolescents. Contradictory, marginally significant inverse trends were also
312
Perla et al.
Table 1. Demographic Characteristics and Asthma Outcomes of All Participants 6–15 Years of Age, and by Pesticide Data Set; NHANES 1999–2008 1999–2004 12–15 years of age
1999–2008 6–15 years of age All NHANES (N = 10,077)
Archives of Environmental & Occupational Health 2015.70:309-322.
Characteristic Age∗ 6–11 12–15 Sex∗ Female Male Race∗ Non-Hispanic White Mexican American Non-Hispanic Black Country of birth Mexico/Other USA Missing/refused BMI—for age∗∗ Underweight Average Overweight Obese Missing/refused Poverty index ratio (PIR)∗ 75th percentile 2. 12–15 years of age
30 32 43
1.00 Referent 1.28 (0.78–2.12) 1.11 (0.64–1.94)
1.00 Referent 1.00 Referent 1.22 (0.61–2.46) 1.34 (0.70–2.58) 1.53 (0.70–3.32) 1.69 (0.81–3.51) 1.19 (0.53–2.65) 1.32 (0.61–2.82)
ADJc PRR
Current Asthma
Prevalence rate ratio (95% CI)
Casesa UNADJ PRRb 38 34 34 40
ADJc PRR 1.00 Referent 0.99 (0.54–1.80) 1.47 (0.77–2.78) 1.13 (0.63–2.05)
Ever Asthma
DMAP creatinine corrected (nmol/L) (N = 1,486) 75th percentile 168–6,115 367 62 1.03 (0.54–1.95) DEAP creatinine corrected (nmol/L) (N = 2,404)
ISSN: 1933-8244 (Print) 2154-4700 (Online) Journal homepage: http://www.tandfonline.com/loi/vaeh20
Biomarkers of Insecticide Exposure and Asthma in Children: A National Health and Nutrition Examination Survey (NHANES) 1999–2008 Analysis M. E. Perla, Tessa Rue, Allen Cheadle, James Krieger & Catherine J. Karr To cite this article: M. E. Perla, Tessa Rue, Allen Cheadle, James Krieger & Catherine J. Karr (2015) Biomarkers of Insecticide Exposure and Asthma in Children: A National Health and Nutrition Examination Survey (NHANES) 1999–2008 Analysis, Archives of Environmental & Occupational Health, 70:6, 309-322, DOI: 10.1080/19338244.2014.910490 To link to this article: http://dx.doi.org/10.1080/19338244.2014.910490
Accepted author version posted online: 22 Aug 2014.
Submit your article to this journal
Article views: 177
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=vaeh20 Download by: [Australian National University]
Date: 07 November 2015, At: 22:14
Archives of Environmental & Occupational Health (2015) 70, 309–322 C Taylor & Francis Group, LLC Copyright ISSN: 1933-8244 print / 2154-4700 online DOI: 10.1080/19338244.2014.910490
Biomarkers of Insecticide Exposure and Asthma in Children: A National Health and Nutrition Examination Survey (NHANES) 1999–2008 Analysis M. E. PERLA1, TESSA RUE2, ALLEN CHEADLE3, JAMES KRIEGER4, and CATHERINE J. KARR5
Archives of Environmental & Occupational Health 2015.70:309-322.
1
Northwest Pediatric Environmental Health Specialty Unit, School of Public Health, University of Washington, Seattle, Washington, USA 2 The Institute of Translational Health Sciences, University of Washington, Seattle, Washington, USA 3 Department of Biostatisticis, Center for Biomedical Statistics, School of Public Health, University of Washington, Seattle, Washington, USA 4 Public Health Department, Seattle and King County, Seattle, Washington, USA 5 Department of Pediatrics, University of Washington, Seattle, Washington, USA Received 29 October 2013, Accepted 21 March 2014
Pesticide exposure is a potential risk factor for increased asthma prevalence among children. The authors used National Health and Nutrition Examination Survey (1999–2008) biomarker data to evaluate dialkylphosphate (DAP) urinary concentrations, serum dichlorodiphenyldichloroethylene (DDE), and asthma among school-aged children (Mexican American, Non-Hispanic Black, NonHispanic White). Poisson logistic regression included age, sex, nativity, poverty index ratio, tobacco smoke exposure, and body mass index covariates. No association was found between DAP (N = 2,777) and asthma outcomes; adverse effect of DDE (N = 940) was suggested for Current Wheeze. Subgroup analyses identified positive associations with some asthma outcomes among Non-Hispanic Blacks, whereas inverse associations were identified among Mexican Americans. Results support previous associations observed among children’s DDE exposure and wheeze. Characterization of risk factors for pesticide exposure and disease recognition among Mexican Americans is needed. Keywords: asthma, children, DDE, epidemiology, pesticides
Introduction There has been an increase in the asthma prevalence rate among US children during the past 30 years.1–3 Children have a higher asthma prevalence rate than adults, and low-income racial and ethnic groups continue to be disproportionately affected.4–6 The causes of asthma and the reasons for its increased occurrence have remained little understood7; however, a number of well-established environmental risk factors have been identified for its exacerbation in individuals. The role of nonpersistent and bioaccumulative pesticide exposures in the development of asthma and the exacerbation of this disease has received attention due to the widespread and continued use of pesticides in both urban8–12 and rural13,14 environments and biological plausibility.15 Few epidemiological studies have explored the relationship between specific pesticide chemical types and asthma or asthma-related symptoms
in children. Prospective studies have found associations between prenatal and breast milk exposures to organochlorine pesticides and asthma, wheeze, and lower respiratory tract infections in newborn and young children in Spain.16–19 In the Ontario Farm Family Health Study, researchers did not observe clear associations between self-reported maternal pesticide exposures during pregnancy and asthma outcomes of their offspring.20 Studies of adults in agricultural settings, in contrast, have identified links between self-reported lifetime or current exposure to pesticides and respiratory health.21–23 We explored the hypothesis that pesticides are associated with asthma in school-aged children 7–16 years of age using population-based data of biomarker concentrations. Given the disparities in asthma outcomes and exposure to pesticides among subpopulations, effects by country of birth and race/ethnicity were also examined.
Methods Address correspondence to M. E. Perla, PhD, MS, Northwest Pediatric Environmental Health Specialty Unit, School of Public Health, University of Washington, Box 359739, 325 9th Avenue, Seattle, WA 98104, USA. E-mail: [email protected]
Data Source We obtained demographic, asthma, early childhood health, and chemical biomarker data collected by the National
310 Health and Nutritional Examination Surveys (NHANES). NHANES is conducted following a complex, multistage area sampling protocol to generate cross-sectional representative data of the civilian noninstitutionalized population living in the United States. To increase sample sizes, we merged several NHANES survey cycle data sets from 1999 to 2008. During the greater part of this time frame, Mexican Americans, non-Hispanic African Americans, low-income Non-Hispanic White Americans, and adolescents were oversampled for the purpose of generating sufficient sample sizes to derive reliable population-based estimates.24
Archives of Environmental & Occupational Health 2015.70:309-322.
Asthma and Asthma Symptoms Recognizing that the likelihood of a clinical diagnosis or selfrecognition of asthma may vary among subpopulations,25–27 we examined 3 dichotomous asthma outcome variables derived from NHANES medical examination and respiratory health survey questionnaires: (1) “Has a doctor or other health professional ever told you that you have asthma?” (Ever Asthma); (2) “Do you still have asthma?” (Current Asthma); and (3) “In the past 12 months have you had wheezing or whistling in your chest?” (Current Wheeze).
Biomarkers of Exposure The primary independent variables of interest were (1) common metabolites of organophosphate insecticides (dialkyl phosphates), a nonpersistent insecticide, and (2) dichlorodiphenyldichloroethylene (p,p’-DDE), the major metabolite of the organochlorine insecticide dichlorodiphenyltrichloroethane (DDT). NHANES participants 6 years of age and older submitted a whole-blood specimen and spot urine specimen. From the blood specimen, cotinine serum levels, a metabolite of tobacco smoke exposure and covariate of interest, were measured for all participants. A randomly selected subset of participants 6 years and older had urine specimens evaluated for concentrations of the 6 possible dialkylphosphate (DAP) metabolites of organophosphate pesticides: dimethylphosphate (DMP), dimethylthiophosphate (DMTP), dimethyldithiophosphate (DMDTP), diethylphosphate (DEP), diethylthiophosphate (DETP), and diethyldithiophosphate (DEDTP). For the analysis of DAP, we summed the 6 organophosphate metabolite concentrations after converting metabolites to nanomolar units by dividing the analyte concentrations by their molecular weight and multiplying by 1000. We adjusted the estimates of urinary biomarkers by correcting each organophosphate metabolite for urinary creatinine before conversion and DAP aggregation.28,29 Because specific organophosphate urinary metabolites had different percentages of observations above the limit of detection, ranging from 29% to 70%, we summed dimethyl alkylphosphate pesticides and diethyl alkylphosphate separately to yield DMAP and DEAP concentrations, for sensitivity analyses. For a separate, randomly selected subset of participants 12 years of age and older, DDE was measured in serum.
Perla et al. Nearly 100% of the DDE observations were above the limit of detection.
Covariates Our a priori selection of available NHANES covariates for descriptive statistics focused on risk factors associated with pesticide exposure and asthma or its diagnosis. These covariates included demographics and socioeconomic status (age, sex, race/ethnicity, country of birth, poverty index ratio [PIR], health insurance coverage), body mass index (BMI) by age, and exposure to tobacco smoke (prenatal smoking, cotinine). We examined data of participants 6–11 and 12–15 years of age available in the DAP data set, and data of adolescents 12–15 years of age available in the DDE data set. To evaluate disparities by race/ethnicity, we conducted stratified analyses of Non-Hispanic Whites, Mexican Americans, and NonHispanic Blacks. Although NHANES collects data from other racial/ethnic groups including Other Latinos (who do not have Mexican American heritage but who identify as Hispanic or Latino), these groups could not be considered for analysis due to sample size considerations. We dichotomized country of birth (US-born, foreign-born) and health insurance coverage (yes, no). The PIR was dichotomized to generally represent families meeting or not meeting poverty eligibility guidelines for federal programs: ie, PIR 10%) of this cross-sectional study prompted use of Poisson logistic regression with robust variance estimation.31 To select exposure percentile cutpoints reflective of changes in risk of asthma with increasing pesticide exposure levels, we used linear splines to model the shape of the dose-response trend.32 Ever Asthma had the highest prevalence rate and overlaps with Current Asthma and Current Wheeze; therefore, cutpoints identified for Ever Asthma were used to evaluate all health outcomes. This process identified 25th percentile as cutpoint for creatinine-corrected DAP and the 40th and 80th percentiles as biomarker cutpoints for DDE. We used these cutpoints to define levels of exposure for each pesticide.
Archives of Environmental & Occupational Health 2015.70:309-322.
Archives of Environmental & Occupational Health We estimated the unadjusted and adjusted asthma prevalence ratios for increasing exposure levels using the 25th percentile for DAP and the 40th percentile for DDE as the referent categories. Adjusted prevalence rate ratio (PRR) models for the DAP analysis included sex, race/ethnicity, BMI (continuous), PIR prenatal smoking, and log cotinine (continuous). As the DDE analysis had a smaller sample sizes and number of cases, for these models we reduced the number of covariates and adjusted for sex, BMI and the poverty index ratio. Due to the smaller race/ethnicity-stratified sample sizes, particularly for the DDE data set, we further reduced the number of covariates by using bivariate regression to select variables with the greatest influence in our sample. The stratified DAP models were adjusted for sex, cotinine, and BMI; the stratified analysis of DDE included BMI as a covariate. Statistical tests were 2-sided and our threshold for statistical significance was p < .025 and for marginal significance .025 ≤ p ≤ .10. The geometric mean of DAP in the 1999–2001 cohort of participants was significantly higher than in the larger analytical sample, so we evaluated the influence of results specific to survey cycle. We assessed the impact of tobacco smoke exposure on our results by removing participants with high cotinine exposure levels indicative of an active smoker.33 In addition, for DAP metabolites, we evaluated the potential impact of fasting time, which may influence the individual measurements of creatinine. Because we could not include country of birth in the stratified multivariate analysis, we conducted separate analyses restricted to US-born participants. This effort focused on Mexican Americans, since they had the largest percentage of foreign-born participants in the DAP (20%) and the DDE (25%) pesticide data sets, whereas Non-Hispanic Blacks and Non-Hispanic Whites had roughly 3% foreign-born in both pesticide data sets. For these analyses, only BMI was included as a covariate. We used Stata 1034 for our analysis. NHANES survey sample weights were used in estimation of unadjusted and adjusted geometric means, prevalence rate ratios, standard error estimates, and 95% confidence intervals, as well as exposure percentiles.
Results NHANES participants from 1999 to 2008 included 10,077 6–15-year-old Non-Hispanic Whites, Mexican Americans, and Non-Hispanic Blacks (Table 1). Thirty-six percent (3,622) 6–15-year-old participants submitted urine samples for measurement of DAP concentrations. With the exception of oversampling differences of adolescents, individuals who contributed urine samples for DAP analysis were similar to the larger NHANES participant group. During the 1999–2004 time frame, 962 participants aged 12–15 years were randomly selected to submit a blood sample for evaluation of DDE. For the analysis of the relationship between pesticides and asthma, the DAP data set was reduced by 845 (23%), primarily due to the incomplete analysis of all 6 dialkylphosphate metabolites needed to generate a DAP vari-
311 able. The DDE sample size was reduced by 22 (2%), primarily due to missing prenatal smoking and BMI observations. In both pesticide data sets, Non-Hispanic Blacks had the highest prevalence rates of Ever Asthma and Current Asthma compared with Non-Hispanic Whites and Mexican Americans (Table 2). With the exception of Current Wheeze in the DDE data set, Non-Hispanic Blanks also had the highest estimates of Current Wheeze. There were some statistical differences of estimated uncorrected geometric mean DAP pesticide concentrations by race ethnicity; however, there were no differences in the creatine-corrected estimates (Table 2). The DDE geometric mean among Mexican Americans was roughly 3 times higher than those of Non-Hispanic Whites and Non-Hispanic Blacks. Mexican Americans also had the lowest geometric mean levels of cotinine in both data sets: NonHispanic Whites had cotinine levels roughly 3 times higher than Mexican Americans, and Non-Hispanic Blacks had estimates 4 times higher. Unadjusted and adjusted asthma outcome prevalence rate ratios by increasing urinary DAP concentration percentiles were largely null, with point estimates generally greater than 1.0 for all participants (Tables 3a and 3b). Sensitivity analyses of DMAP and DEAP revealed similar null results (data in Appendices A and B). In stratified analysis, among NonHispanic Blacks, a positive trend was suggested for Current Wheeze among 6–11-year-old and Ever and Current Asthma among 12–15-year-olds. In contrast, an inverse trend was suggested for Ever and Current Asthma and Current Wheeze among Mexican American adolescents. Increasing percentiles of serum DDE were associated with increased Current Wheeze but not Ever or Current Asthma (Table 4). Subgroup analyses were limited by small sample size but supported this association among Non-Hispanic Whites and Non-Hispanic Blacks. In contrast, an inverse effect among Mexican Americans was shown for both Current Asthma and Current Wheeze. Our sensitivity analyses addressing survey cycle influence, number of hours fasted since specimen collection, and removing individuals with high body burden of cotinine did not alter the patterns reported in Tables 3a and 3b. Additional sensitivity analyses focused on US-born Mexican Americans. Increasing DAP and of DDE exposure levels did not differ from our core analyses (data available in Appendices A and B).
Comment Overall, this study revealed no clear associations between asthma and biomarkers of DAP or DDT insecticides among school-aged children in the United States. Although the analyses of asthma outcomes and urinary metabolites of DAP were largely null overall, subgroup analyses demonstrated marginally significant positive trends for Non-Hispanic Blacks for Current Wheeze among 6–11-yearolds, and Ever and Current Asthma among adolescents. Contradictory, marginally significant inverse trends were also
312
Perla et al.
Table 1. Demographic Characteristics and Asthma Outcomes of All Participants 6–15 Years of Age, and by Pesticide Data Set; NHANES 1999–2008 1999–2004 12–15 years of age
1999–2008 6–15 years of age All NHANES (N = 10,077)
Archives of Environmental & Occupational Health 2015.70:309-322.
Characteristic Age∗ 6–11 12–15 Sex∗ Female Male Race∗ Non-Hispanic White Mexican American Non-Hispanic Black Country of birth Mexico/Other USA Missing/refused BMI—for age∗∗ Underweight Average Overweight Obese Missing/refused Poverty index ratio (PIR)∗ 75th percentile 2. 12–15 years of age
30 32 43
1.00 Referent 1.28 (0.78–2.12) 1.11 (0.64–1.94)
1.00 Referent 1.00 Referent 1.22 (0.61–2.46) 1.34 (0.70–2.58) 1.53 (0.70–3.32) 1.69 (0.81–3.51) 1.19 (0.53–2.65) 1.32 (0.61–2.82)
ADJc PRR
Current Asthma
Prevalence rate ratio (95% CI)
Casesa UNADJ PRRb 38 34 34 40
ADJc PRR 1.00 Referent 0.99 (0.54–1.80) 1.47 (0.77–2.78) 1.13 (0.63–2.05)
Ever Asthma
DMAP creatinine corrected (nmol/L) (N = 1,486) 75th percentile 168–6,115 367 62 1.03 (0.54–1.95) DEAP creatinine corrected (nmol/L) (N = 2,404)
